BMS-02 — BMS Architecture

ASSUMPTION — The block names in this lesson reflect general industry practice. A real OEM may name these functions differently or combine them into different physical hardware; no real OEM architecture or internal design is used here.

1. Where Do You Put the Measurement and the Control?

Picture a 96-cell pack: measuring every cell’s voltage takes at least 97 measurement points. Do you route all of them to one central board, or spread small electronic units close to the cells? That single decision drives wiring length, cost, noise immunity, and serviceability — which is why architecture is one of the most fundamental decisions in BMS design.

2. Three Core Building Blocks

Component	Job	Location
CMU (Cell Monitoring Unit)	Measures cell voltage/temperature, and in most designs also houses the passive balancing resistors	Closest point to the cells
BMU (Battery Management Unit)	Computes SOC/SOH/SOP, makes protection and fault decisions, controls contactors, talks to the VCU	The BMS’s “brain”
BJB (Battery Junction Box)	Houses the main contactors, pre-charge circuit, fuse, current sensor, and HVIL	The HV switching point

The heart of a CMU is usually an AFE (Analog Front End) / Cell Monitoring IC that can monitor 6 to 18 cells (details in BMS-03) — sitting close to the cells shortens the sense wires, which reduces noise exposure. The BMU runs a layered software stack: the state machine (BMS-14), SOC/SOH/SOP algorithms (BMS-08/09/10), fault management (BMS-13), and communications (BMS-15) — usually on an automotive-grade MCU, under real-time constraints. Grouping the BJB into its own box has a simple reason: keeping the HV switching elements in one protected place both restricts access and makes servicing easier.

3. Which Way Signals Flow

Cell → CMU → BMU → VCU

Measurement data flows bottom-up, control commands flow top-down. A CMU never issues a command directly to a contactor — that decision is always made in the BMU.

ASSUMPTION — In some real implementations, the BJB’s control logic may be integrated into the BMU, or the BJB may have its own small controller. This lesson keeps the functions separate to clarify the conceptual distinction.

4. Three Architectural Approaches

Centralized: all measurement is concentrated on a single board, with cell-voltage wires routed there directly. Fewer electronic units, lower cost (in small packs) — but for 96 cells, routing more than 96 individual voltage lines gets physically unwieldy, the wiring harness bulks up, and the design becomes more vulnerable to EMI.

Distributed: measurement is split across small CMUs placed close to the cells; the CMUs connect to each other and to the BMU, typically over a daisy-chain link. Short wire runs, good EMI behavior — but more electronic units, and a break anywhere in the chain can take out the data from every CMU downstream of it.

Modular: the pack is divided into modules, each with its own measurement electronics. Think of it as the distributed approach aligned to module boundaries — it lets you service a faulty module by replacing just that module instead of the entire pack.

Criterion	Centralized	Distributed	Modular
Measurement wiring	Long/extensive	Short/minimal	Short/minimal
EMI sensitivity	High	Low	Low
Cost	Low (small pack)	Medium	Medium-high
Scalability	Low	High	High
Serviceability	Medium	Medium	High (module swap)

There’s no single “best” architecture — the choice depends on pack size, cost target, safety requirements, and manufacturing strategy.

In distributed/modular architectures, the BMU is usually master and the CMUs are slaves: the master sends measurement commands, collects data, keeps everything synchronized, and monitors whether each slave is responding (chain integrity).

5. How Architecture Gets Verified in Production

End-of-Line (EOL) testing confirms that every CMU is correctly addressed, the daisy-chain is intact, and communication latency stays within budget (BMS-20). HIL testing simulates scenarios like a CMU losing communication, to confirm the BMU reacts correctly — for example, by flagging the affected cells as “no data” and staying on the safe side (BMS-19).

6. How It Connects to Other Systems

Through the contactors in the BJB, the BMS interacts with the inverter, the charger, and the DC/DC converter — it won’t connect these loads to HV until safe conditions are met. It requests cooling/heating from thermal management (BMS-12) and reports SOC/SOH/SOP to the VCU (BMS-15). Architecture choice also affects functional safety: in a distributed architecture, local checks each CMU can perform on its own (like a local overvoltage threshold) can act as an extra safety layer independent of the central BMU (BMS-17).

Summary

• The CMU measures, the BMU decides, the BJB switches.
• Centralized/Distributed/Modular each trade off differently; the choice depends on pack size, EMI, cost, and serviceability.
• Measurement flows up, control flows down — CMUs never command contactors directly.
• Architecture gets verified in production through EOL testing and HIL scenarios.

Sources

• Gregory L. Plett, Battery Management Systems, Volume I & II — architecture and measurement concepts.
• ISO 26262 — functional safety architecture context.